clinical article J Neurosurg 122:363–372, 2015
Gamma Knife radiosurgery of large skull base meningiomas Robert M. Starke, MD, MSc,1 Colin J. Przybylowski, BS,2 Mukherjee Sugoto, MD,3 Francis Fezeu, MD, PhD,1 Ahmed J. Awad, MD,1 Dale Ding, MD,1 James H. Nguyen, MD,1 and Jason P. Sheehan, MD, PhD1 Department of Neurosurgery, 2School of Medicine, and 3Neuroradiology, University of Virginia Health System, Charlottesville, Virginia 1
Object Stereotactic radiosurgery (SRS) has become a common treatment modality for intracranial meningiomas. Skull base meningiomas greater than 8 cm3 in volume have been found to have worse outcomes following SRS. When symptomatic, patients with these tumors are often initially treated with resection. For tumors located in close proximity to eloquent structures or in patients unwilling or unable to undergo a resection, SRS may be an acceptable therapeutic approach. In this study, the authors review the SRS outcomes of skull base meningiomas greater than 8 cm3 in volume, which corresponds to a lesion with an approximate diameter of 2.5 cm. Methods The authors reviewed the data in a prospectively compiled database documenting the outcomes of 469 patients with skull base meningiomas treated with single-session Gamma Knife radiosurgery (GKRS). Seventy-five patients had tumors greater than 8 cm3 in volume, which was defined as a large tumor. All patients had a minimum follow-up of 6 months, but patients were included if they had a complication at any time point. Thirty patients were treated with upfront GKRS, and 45 were treated following microsurgery. Patient and tumor characteristics were assessed to determine predictors of new or worsening neurological function and tumor progression following GKRS. Results After a mean follow-up of 6.5 years (range 0.5–21 years), the tumor volume was unchanged in 37 patients (49%), decreased in 26 patients (35%), and increased in 12 patients (16%). Actuarial rates of progression-free survival at 3, 5, and 10 years were 90.3%, 88.6%, and 77.2%, respectively. Four patients had new or worsened edema following GKRS, but preexisting edema decreased in 3 patients. In Cox multivariable analysis, covariates associated with tumor progression were 1) presentation with any cranial nerve (CN) deficit from III to VI (hazard ratio [HR] 3.78, 95% CI 1.91–7.45; p < 0.001), history of radiotherapy (HR 12.06, 95% CI 2.04–71.27; p = 0.006), and tumor volume greater than 14 cm3 (HR 6.86, 95% CI 0.88–53.36; p = 0.066). In those patients with detailed clinical follow-up (n = 64), neurological function was unchanged in 37 patients (58%), improved in 16 patients (25%), and deteriorated in 11 patients (17%). In multivariate analysis, the factors predictive of new or worsening neurological function were history of surgery (OR 3.00, 95% CI 1.13–7.95; p = 0.027), presentation with any CN deficit from III to VI (OR 3.94, 95% CI 1.49–10.24; p = 0.007), and decreasing maximal dose (OR 0.76, 95% CI 0.63–0.93; p = 0.007). Tumor progression was present in 64% of patients with new or worsening neurological decline. Conclusions Stereotactic radiosurgery affords a reasonable rate of tumor control for large skull base meningiomas and does so with a low incidence of neurological deficits. Those with a tumor less than 14 cm3 in volume and without presenting CN deficit from III to VI were more likely to have effective tumor control. http://thejns.org/doi/abs/10.3171/2014.10.JNS14198
M
Key Words stereotactic radiosurgery; Gamma Knife radiosurgery; meningioma; skull base; outcome; recurrence; microsurgery
are the most frequently reported primary intracranial neoplasms in the US. Most of these lesions are benign (i.e., WHO Grade I) with a low tendency for invasion and recurrence and eningiomas
a natural history of slow growth. Within the intracranial space, there is a narrow capacity for mass expansion. The primary approach for large meningiomas has been resection. Despite a dramatic decline in surgical morbidity for
Abbreviations CN = cranial nerve; CPA = cerebellopontine angle; GKRS = Gamma Knife radiosurgery; HR = hazard ratio; SRS = stereotactic radiosurgery. submitted January 26, 2014. accepted October 14, 2014. include when citing Published online December 5, 2014; DOI: 10.3171/2014.10.JNS14198. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. ©AANS, 2015
J Neurosurg Volume 122 • February 2015
363
R. M. Starke et al.
meningiomas reported over the last 2 decades, gross-total resection of meningiomas in critical locations, especially the skull base, has remained a challenge, and patients often require multiple surgeries leading to increased morbidity and mortality.6,10,36 In recent years, primary or adjuvant treatment with stereotactic radiosurgery (SRS) has gained favor. SRS has demonstrated its safety and efficacy in the control of benign tumors, particularly for small to moderately sized meningiomas. Tumor control rates for WHO Grade I skull base meningiomas after SRS average approximately 91% and 88% at 5 and 10 years, respectively.13,19,22,27,33–35,38,39,48,51 Most meningioma radiosurgical series have excluded large volume tumors. Traditionally, a tumor diameter of 30–35 mm was the recommended cutoff for radiosurgery.30–32 In prior studies, large meningiomas have been treated with varying degrees of success, but radiosurgery outcomes appear to worsen for patients harboring meningiomas greater than 8 cm3 in volume.8,15,16,20,42 An 8-cm3 volume corresponds to a lesion with an approximate diameter of 2.5 cm. Radiosurgery to lesions in excess of 8-cm3 volume has been linked to worsening outcome with arteriovenous malformations, acoustic neuromas, and brain metastases as well.12,18,26 In the present study, we retrospectively reviewed data of patients harboring large WHO Grade I skull base meningiomas (volume > 8 cm3) who were treated with single-session Gamma Knife radiosurgery (GKRS) to identify prognostic factors associated with successful and adverse radiological and clinical outcomes.
Methods
Patient Population This is a retrospective analysis of a prospectively maintained database approved by the University of Virginia institutional review board. The database was assessed from 1989 to 2013 for patients with skull base meningiomas treated with single-session GKRS at the University of Virginia (n = 469). The diagnosis was confirmed either by tissue pathology or characteristic findings for meningiomas on neuroimaging studies. Tumors typically exhibited radiological features of a meningioma including dural base, extraaxial location, uniform contrast enhancement, and intratumoral calcification. Exclusion criteria included patients with multiple meningiomas, history of prior cancer, follow-up less than 6 months unless there was a complication, and a confirmed tumor histological grade greater than WHO Grade I. All patients with tumor progression and all those with complications following GKRS were included, regardless of follow-up duration. Consequently, 75 patients had large tumors (> 8 cm3) as assessed with tumor volumetry using radiosurgical planning software at the time of GKRS and were included for analysis. Patients were considered for GKRS if they were not disabled by their tumor, i.e., Karnofsky Performance Scale score < 70. For those who underwent upfront treatment with GKRS, patients were not candidates for primary resection based upon their advanced age, the projected operative risks based on medical comorbidities, and/or refusal of microsurgical resection. Adjuvant radiosurgery was 364
J Neurosurg Volume 122 • February 2015
performed in patients with recurrence of lesions following microsurgical excision or as part of multimodality treatment whereby the risks of surgical gross-total resection outweighed the benefits of multimodality therapy. Patient and Tumor Attributes Of the 75 patients with large skull base meningiomas, 51 were female (68%) and 24 were male (32%), with a median age of 55 years (range 19–85 years). Forty-five patients (60%) had a history of surgery and histologically confirmed WHO Grade I meningiomas, while the remaining 30 patients treated with upfront radiosurgery had neuroimaging characteristics and a clinical history consistent with a benign meningioma (40%). Alteration in function of cranial nerves (CNs) III, IV, V, or VI was the most common neurological deficit on presentation. The mean volume of tumors prior to radiosurgery was 14.1 ± 6.7 cm3 (median 12.4 cm3, range 8.1–54.8 cm3). The parasellar region (n = 42, 56%) and cerebellopontine angle (CPA) (n = 10, 13%) were the most common tumor locations. Preoperative patient characteristics, presentations, and tumor characteristics are detailed in Table 1. Radiosurgical Technique Our radiosurgical technique has been previously described.49 In brief, patients underwent placement of a stereotactic Leksell model G stereotactic frame (Elekta Instruments Inc.) in the operating room. During frame placement, they received monitored anesthesia administered by an anesthesiologist. Stereotactic MRI was then obtained for the treatment planning. Pre- and postcontrast thin-slice (1-mm) volume acquisition axial and coronal MRI sequences were obtained. When an MRI was not able to be obtained due to medical contraindications (e.g., a cardiac pacemaker), a thin-slice stereotactic CT scan was obtained with and without contrast administration. Radiosurgical dose plans were formulated under the direction of a neurosurgeon in conjunction with a medical physicist and radiation oncologist. In general, the maximum dose to the optic apparatus was kept to 8 Gy or less. The Leksell Gamma Unit Model U was used until July 2001 when the C model (Elekta Instruments, Inc.) was instituted. The Gamma Knife Perfexion unit has been used to treat patients since 2007. The Kula software was used for dose planning from 1990 to June 1994, and then subsequently Elekta’s Gamma Plan software was used. All patients in this series were treated with single-session radiosurgery. Initial tumor volume was assessed by contouring and then using Gamma Plan software. Gamma Knife Radiosurgery Parameters The mean dose to the tumor margin was 13.5 ± 3.5 Gy (range 4.8–30 Gy) with a mean maximal dose of 31 ± 5 Gy (range 23–40 Gy). Only 1 patient received a marginal dose of less than 9 Gy, and the 5th to 95th percentiles for marginal dose ranged from 9 to 20 Gy. The majority of tumors were treated with multiple isocenters (mean 13 ± 10 isocenters, range 1–43 isocenters) to the mean prescription isodose line of 42% ± 9% (range 30%–55%). Radiosurgery parameters are detailed in Table 2.
Radiosurgery of large skull base meningiomas
TABLE 1. Baseline characteristics of 75 patients with large skull base meningiomas treated with GKRS Characteristic
Value (%)
Females Age (yrs) Median ± SD Range Previous resection Previous radiotherapy Previous GKRS Initial presentation* Headache Subjective dizziness Seizure CN III/IV/VI V VII VIII IX/X XI XII Hypopituitarism Body weakness Dysgraphia Cognition Location Parasellar Petroclival Clival Tentorial CPA Petrous Sphenoidal Jugular foramen Tumor volume (cm3) Mean ± SD Range Tumor diameter (cm) Mean maximum Range Edema on pre-GKRS imaging
51 (68) 55 ± 13 19–85 45 (60) 4 (5) 3 (4) 18 (28) 14 (22) 3 (5) 29 (45) 33 (52) 9 (14) 11 (17) 1 (2) 0 (0) 1 (2) 2 (3) 4 (6) 2 (3) 1 (2) 42 (56) 9 (12) 6 (8) 3 (4) 10 (13) 2 (3) 2 (3) 1 (1) 14.1 ± 6.7 8.1–54.8 3.8 ± 0.8 2.5–6 4 (5)
* Clinical presentation data available for 64 patients.
Clinical and Radiological Follow-Up Patients were routinely followed clinically and radiologically every 6 months for the first year, annually until 5 years after radiosurgery, and then every 2 years thereafter. At each follow-up visit, a neurological examination was performed to evaluate for new neurological deficits, and neuroimaging studies were reviewed to assess tumor response by both a neurosurgeon and neuroradiologist. All
TABLE 2. GKRS treatment parameters Characteristic
Mean ± SD (range)
Margin dose (Gy) Maximum dose (Gy) Isodose line (%) No. of isocenters
13.5 ± 3.5 (4.8–30) 31 ± 5 (23–40) 42 ± 9 (30–55) 13 ± 10 (1–43)
follow-up evaluation was performed at the University of Virginia, unless the patient was unable to travel to our institution. In such cases, the referring physician performed the follow-up and provided documentation of the patient’s neurological status, as well as follow-up imaging. All neuroimaging studies were reviewed by a neurosurgeon and a neuroradiologist at the University of Virginia. Lesions were categorized into the following locations defined by the presumed origin of maximal volume: sellar or parasellar, CPA, petrous, clival, petroclival, jugular foramen, and sphenoid.2,3,5,50,54,55 Petroclival meningiomas were defined as tumors whose maximal volume was centered over the region between the petrous apex and the upper two-thirds of the clivus.3,50 All parasellar lesions were in close proximity to the sella with cavernous sinus invasion.54 Imaging outcome was determined by the last available radiological examination by a neurosurgeon and a neuroradiologist.47 A decrease or increase in tumor size was defined as a 15% or greater change in tumor volume as compared with the volume at the time of GKRS.47 Tumors with less than a 15% change from their initial volume were considered radiologically stable. Any patient with tumor progression of greater than 15% was considered a treatment failure even if this stabilized with further GKRS or surgery.47 To make the determination of tumor volume, the tumor was outlined on radiological images and serial volumetric calculations were performed using Image J in all patient imaging studies (NIH).47 The presence of perilesional edema was defined as FLAIR or T2-weighted changes around the tumor. These changes were evaluated by a neuroradiologist (M.S.) who assessed the radiosurgical planning MR images and the follow-up MR images for the presence of baseline edema and an exacerbation of the same after radiosurgery. Statistical Analysis Data are presented as median or mean and range for continuous variables, and as frequency and percentage for categorical variables. Calculations of normality were performed by ladder of powers and assessed graphically. Statistical analyses of categorical variables were conducted using chi-square, Fisher’s exact, and Mantel-Haenszel tests for linear association as appropriate. Statistics of means were performed using an unpaired Student t-test, both with and without equal variance (Levene’s test) as necessary and Wilcoxon rank-sum tests when variables were not normally distributed. The following dependent variables were assessed in univariate and multivariate analysis: tumor-free progression, worsening or new deJ Neurosurg Volume 122 • February 2015
365
R. M. Starke et al.
cline in neurological function, and favorable outcome (no tumor progression or worsening or new decline in neurological function). Kaplan-Meier risk of tumor progression was calculated. Factors predictive of tumor progression (p < 0.20)4 were entered into multivariate Cox regression analysis to assess hazard ratios (HRs). Clinical covariates predicting new or worsening decline in neurological function with a univariate p value < 0.20 were included in multivariable logistic regression analysis. Additionally, clinical covariates predicting unfavorable outcome with a univariate p value < 0.20 were included in multivariable logistic regression analysis. Clinically significant variables and interaction expansion covariates were further assessed in both Cox and logistic multivariable analysis as deemed relevant. Youden indices were calculated to determine cutoffs for the dichotomized continuous variable tumor volume that yielded the optimal discrimination of tumor progression and overall outcome. Those p values ≤ 0.05 were considered statistically significant.
Results
Clinical Outcomes Detailed clinical follow-up data were available for 64 patients. Neurological function was unchanged in 37 patients (58%), improved in 16 patients (25%), and declined in 11 patients (17%). Specific alterations in neurological function are displayed in Table 3. The most common CNs
to have new or worsening function following GKRS were II, III, and V, which each occurred in 4 patients (6%). Additionally, the most likely CNs to demonstrate improvement following GKRS were III, V, and VII occurring in 3 (5%), 4 (6%), and 3 (5%) of patients, respectively. In the current series, there was no evidence of vascular injury or brainstem ischemia as a result of SRS. Predictors of new or worsening decline in neurological function in univariate analysis are demonstrated in Table 4. In multivariate analysis, the factors predictive of new or worsening decline in neurological function were history of surgery (OR = 3.00, 95% CI 1.13–7.95; p = 0.027), presentation with any CN deficit from III to VI (OR = 3.94, 95% CI 1.49–10.24; p = 0.007), and decreasing maximal dose (OR = 0.76, 95% CI 0.63–0.93; p = 0.007). Tumor progression was present in 7 (64%) of 11 patients with new or worsening decline in neurological function. Radiological Outcome The mean follow-up duration was 6.5 years (range 0.5– 21 years). During this time, 37 patients (49%) displayed no change in tumor volume, 26 (35%) had decreased volume, and 12 (16%) displayed increased volume. Kaplan-Meier analysis demonstrated radiological progression-free survival at 3, 5, and 10 years to be 90.3%, 88.6%, and 77.2%, respectively (Fig. 1). Four patients had a history of edema prior to SRS (5.3%). Following radiosurgery, 4 patients had new or
TABLE 3. Neurological outcomes for 64 patients with large skull base meningiomas treated with GKRS Variable
Deficit Before GKRS
Total patients (%) Total no. of deficits per patient (%) Dizziness Body weakness Seizure Hypopituitarism Dysgraphia Cognition CN deficit* CN I CN II CN III CN IV CN V CN VI CN VII CN VIII CN IX CN X CN XI CN XII
59 (92)* 14 (22) 4 (6) 3 (5) 2 (3) 2 (3) 1 (2) 52 (81) 2 (3) 19 (30) 24 (38) 3 (5) 33 (52) 16 (25) 9 (14) 11 (17) 0 1 (2) 0 1 (2)
* Some patients had multiple deficits † Includes patients who did not present with a deficit. 366
J Neurosurg Volume 122 • February 2015
Stable Deficit
Improved Deficit
Worsened or New Deficit
37 (58)†
16 (25)
11 (17)
12 (19) 3 (5) 3 (5) 2 (3) 2 (3) 0 33 (52) 2 (3) 15 (23) 18 (28) 3 (5) 26 (41) 16 (25) 6 (9) 10 (16) 0 0 0 1 (2)
2 (3) 1 (2) 0 0 0 1 (2) 12 (19) 0 1 (2) 3 (5) 0 4 (6) 0 3 (5) 1 (2) 0 1 (2) 0 0
0 0 1 (2) 0 0 1 (2) 9 (14) 0 4 (6) 4 (6) 1 (2) 4 (6) 1 (2) 0 0 0 0 0 0
Radiosurgery of large skull base meningiomas
TABLE 4. Predictors of new or worsening neurological deficits Predictive Factor Univariate analysis* Location All other locations Parasellar/petroclival History of radiotherapy History of surgery Any CN deficit from III to VI Seizure Increasing volume Decreasing peripheral dose Decreasing maximal dose Multivariate analysis History of surgery Any CN deficit from III to VI Decreasing maximal dose
OR
95% CI
p Value
1.00 7.10 6.25 2.40 2.98 11.56 1.06 0.82 0.83
0.85–59.54 1.07–36.54 1.08–5.34 1.43–6.23 0.95–141.13 0.98–1.15 0.64–1.05 0.71–0.96
0.071 0.042 0.032 0.004 0.055 0.171 0.118 0.012
3.00 3.94 0.76
1.13–7.95 1.49–10.24 0.63–0.93
0.027 0.007 0.007
* Factors predictive of new/worsening symptoms (p < 0.20).
worsening edema within 18 months of treatment: 1 patient had progression of preexisting edema and 3 patients developed edema following SRS. The edema improved in 1 patient following a short course of steroids but persisted in the other 3 patients. Three patients with preexisting edema experienced improvement in the edema following radiosurgery. There were no cases of malignant transformation of an existing meningioma or radiation-induced secondary tumor development. Factors predictive of tumor progression in univariate analysis are displayed in Table 5. Prior resection was not predictive of tumor control (Fig. 2). Regarding tumor control, Youden indices demonstrated that the optimal breakpoint for tumor volume occurred at 14 cm3 (Fig. 3). In the Cox multivariable analysis, the covariates associated with tumor progression included presentation with any CN deficit from III to VI (HR = 3.78, 95% CI 1.91–7.45; p < 0.001), history of prior radiation therapy (HR = 12.06, 95% CI 2.04–71.27; p = 0.006), and tumor volume greater than 14 cm3 (HR = 6.86, 95% CI 0.88–53.36; p = 0.066; Table 5; Figs. 4 and 5). Further Treatments Following SRS During follow-up, 2 patients underwent fractionated radiation therapy for out-of-field tumor progression (3%), and 5 patients underwent resection for tumor progression (7%). All tumors resected after radiosurgery were confirmed to be WHO Grade I meningiomas. Three patients had evidence of radiological progression of ventriculomegaly, but none of these patients required a CSF diversion procedure. Overall Outcome After SRS Unfavorable outcome, defined as tumor progression and/or new or worsening decline in neurological function, was observed in 12 patients (18.8%), compared with 52 patients (81.2%) who had a favorable outcome. Of those
Fig. 1. Graph of tumor-free progression after GKRS.
with an unfavorable outcome, 7 patients (10.9%) exhibited both tumor progression and neurological decline. Univariate predictors of unfavorable overall outcome (i.e., tumor growth and/or neurological decline) are detailed in Table 6. Statistically significant factors in univariate analysis for an unfavorable outcome included petroclival/parasellar location, prior radiotherapy, prior surgery, any CN deficit from III to VI, diplopia on presentation, preexisting seizures, pre-GKRS edema, and decreasing maximal and peripheral dose (Table 6). Patients with a history of resection before radiosurgery were significantly more likely to have an unfavorable outcome in univariate analysis (OR = 2.37, 95% CI 1.08–5.21; p = 0.032), but prior resection was not an independent predictor of outcome. Youden indices did not provide a significant cutoff for the dichotomized continuous variable “tumor volume” that yielded the optimal discrimination of overall outcome. In the multivariate analysis, the factors predictive of unfavorable outcome included history of prior radiation TABLE 5. Predictors of tumor progression Predictive Factor Univariate analysis* Location All other locations Parasellar/petroclival History of radiotherapy History of hydrocephalus Any CN deficit from III to VI Diplopia on presentation Seizure Tumor volume >14 cm3 Decreasing maximal dose Multivariate analysis History of radiotherapy Any CN deficit from III to VI Tumor volume >14 cm3
HR
95% CI
p Value
1.00 5.11 3.73 1.19 3.29 2.47 7.64 6.86 0.93
0.65–39.91 1.00–13.94 1.00–1.41 1.79–6.05 0.79–7.78 1.53–38.04 0.88–53.36 0.85–1.02
0.120 0.050 0.056
Gamma Knife radiosurgery of large skull base meningiomas Robert M. Starke, MD, MSc,1 Colin J. Przybylowski, BS,2 Mukherjee Sugoto, MD,3 Francis Fezeu, MD, PhD,1 Ahmed J. Awad, MD,1 Dale Ding, MD,1 James H. Nguyen, MD,1 and Jason P. Sheehan, MD, PhD1 Department of Neurosurgery, 2School of Medicine, and 3Neuroradiology, University of Virginia Health System, Charlottesville, Virginia 1
Object Stereotactic radiosurgery (SRS) has become a common treatment modality for intracranial meningiomas. Skull base meningiomas greater than 8 cm3 in volume have been found to have worse outcomes following SRS. When symptomatic, patients with these tumors are often initially treated with resection. For tumors located in close proximity to eloquent structures or in patients unwilling or unable to undergo a resection, SRS may be an acceptable therapeutic approach. In this study, the authors review the SRS outcomes of skull base meningiomas greater than 8 cm3 in volume, which corresponds to a lesion with an approximate diameter of 2.5 cm. Methods The authors reviewed the data in a prospectively compiled database documenting the outcomes of 469 patients with skull base meningiomas treated with single-session Gamma Knife radiosurgery (GKRS). Seventy-five patients had tumors greater than 8 cm3 in volume, which was defined as a large tumor. All patients had a minimum follow-up of 6 months, but patients were included if they had a complication at any time point. Thirty patients were treated with upfront GKRS, and 45 were treated following microsurgery. Patient and tumor characteristics were assessed to determine predictors of new or worsening neurological function and tumor progression following GKRS. Results After a mean follow-up of 6.5 years (range 0.5–21 years), the tumor volume was unchanged in 37 patients (49%), decreased in 26 patients (35%), and increased in 12 patients (16%). Actuarial rates of progression-free survival at 3, 5, and 10 years were 90.3%, 88.6%, and 77.2%, respectively. Four patients had new or worsened edema following GKRS, but preexisting edema decreased in 3 patients. In Cox multivariable analysis, covariates associated with tumor progression were 1) presentation with any cranial nerve (CN) deficit from III to VI (hazard ratio [HR] 3.78, 95% CI 1.91–7.45; p < 0.001), history of radiotherapy (HR 12.06, 95% CI 2.04–71.27; p = 0.006), and tumor volume greater than 14 cm3 (HR 6.86, 95% CI 0.88–53.36; p = 0.066). In those patients with detailed clinical follow-up (n = 64), neurological function was unchanged in 37 patients (58%), improved in 16 patients (25%), and deteriorated in 11 patients (17%). In multivariate analysis, the factors predictive of new or worsening neurological function were history of surgery (OR 3.00, 95% CI 1.13–7.95; p = 0.027), presentation with any CN deficit from III to VI (OR 3.94, 95% CI 1.49–10.24; p = 0.007), and decreasing maximal dose (OR 0.76, 95% CI 0.63–0.93; p = 0.007). Tumor progression was present in 64% of patients with new or worsening neurological decline. Conclusions Stereotactic radiosurgery affords a reasonable rate of tumor control for large skull base meningiomas and does so with a low incidence of neurological deficits. Those with a tumor less than 14 cm3 in volume and without presenting CN deficit from III to VI were more likely to have effective tumor control. http://thejns.org/doi/abs/10.3171/2014.10.JNS14198
M
Key Words stereotactic radiosurgery; Gamma Knife radiosurgery; meningioma; skull base; outcome; recurrence; microsurgery
are the most frequently reported primary intracranial neoplasms in the US. Most of these lesions are benign (i.e., WHO Grade I) with a low tendency for invasion and recurrence and eningiomas
a natural history of slow growth. Within the intracranial space, there is a narrow capacity for mass expansion. The primary approach for large meningiomas has been resection. Despite a dramatic decline in surgical morbidity for
Abbreviations CN = cranial nerve; CPA = cerebellopontine angle; GKRS = Gamma Knife radiosurgery; HR = hazard ratio; SRS = stereotactic radiosurgery. submitted January 26, 2014. accepted October 14, 2014. include when citing Published online December 5, 2014; DOI: 10.3171/2014.10.JNS14198. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. ©AANS, 2015
J Neurosurg Volume 122 • February 2015
363
R. M. Starke et al.
meningiomas reported over the last 2 decades, gross-total resection of meningiomas in critical locations, especially the skull base, has remained a challenge, and patients often require multiple surgeries leading to increased morbidity and mortality.6,10,36 In recent years, primary or adjuvant treatment with stereotactic radiosurgery (SRS) has gained favor. SRS has demonstrated its safety and efficacy in the control of benign tumors, particularly for small to moderately sized meningiomas. Tumor control rates for WHO Grade I skull base meningiomas after SRS average approximately 91% and 88% at 5 and 10 years, respectively.13,19,22,27,33–35,38,39,48,51 Most meningioma radiosurgical series have excluded large volume tumors. Traditionally, a tumor diameter of 30–35 mm was the recommended cutoff for radiosurgery.30–32 In prior studies, large meningiomas have been treated with varying degrees of success, but radiosurgery outcomes appear to worsen for patients harboring meningiomas greater than 8 cm3 in volume.8,15,16,20,42 An 8-cm3 volume corresponds to a lesion with an approximate diameter of 2.5 cm. Radiosurgery to lesions in excess of 8-cm3 volume has been linked to worsening outcome with arteriovenous malformations, acoustic neuromas, and brain metastases as well.12,18,26 In the present study, we retrospectively reviewed data of patients harboring large WHO Grade I skull base meningiomas (volume > 8 cm3) who were treated with single-session Gamma Knife radiosurgery (GKRS) to identify prognostic factors associated with successful and adverse radiological and clinical outcomes.
Methods
Patient Population This is a retrospective analysis of a prospectively maintained database approved by the University of Virginia institutional review board. The database was assessed from 1989 to 2013 for patients with skull base meningiomas treated with single-session GKRS at the University of Virginia (n = 469). The diagnosis was confirmed either by tissue pathology or characteristic findings for meningiomas on neuroimaging studies. Tumors typically exhibited radiological features of a meningioma including dural base, extraaxial location, uniform contrast enhancement, and intratumoral calcification. Exclusion criteria included patients with multiple meningiomas, history of prior cancer, follow-up less than 6 months unless there was a complication, and a confirmed tumor histological grade greater than WHO Grade I. All patients with tumor progression and all those with complications following GKRS were included, regardless of follow-up duration. Consequently, 75 patients had large tumors (> 8 cm3) as assessed with tumor volumetry using radiosurgical planning software at the time of GKRS and were included for analysis. Patients were considered for GKRS if they were not disabled by their tumor, i.e., Karnofsky Performance Scale score < 70. For those who underwent upfront treatment with GKRS, patients were not candidates for primary resection based upon their advanced age, the projected operative risks based on medical comorbidities, and/or refusal of microsurgical resection. Adjuvant radiosurgery was 364
J Neurosurg Volume 122 • February 2015
performed in patients with recurrence of lesions following microsurgical excision or as part of multimodality treatment whereby the risks of surgical gross-total resection outweighed the benefits of multimodality therapy. Patient and Tumor Attributes Of the 75 patients with large skull base meningiomas, 51 were female (68%) and 24 were male (32%), with a median age of 55 years (range 19–85 years). Forty-five patients (60%) had a history of surgery and histologically confirmed WHO Grade I meningiomas, while the remaining 30 patients treated with upfront radiosurgery had neuroimaging characteristics and a clinical history consistent with a benign meningioma (40%). Alteration in function of cranial nerves (CNs) III, IV, V, or VI was the most common neurological deficit on presentation. The mean volume of tumors prior to radiosurgery was 14.1 ± 6.7 cm3 (median 12.4 cm3, range 8.1–54.8 cm3). The parasellar region (n = 42, 56%) and cerebellopontine angle (CPA) (n = 10, 13%) were the most common tumor locations. Preoperative patient characteristics, presentations, and tumor characteristics are detailed in Table 1. Radiosurgical Technique Our radiosurgical technique has been previously described.49 In brief, patients underwent placement of a stereotactic Leksell model G stereotactic frame (Elekta Instruments Inc.) in the operating room. During frame placement, they received monitored anesthesia administered by an anesthesiologist. Stereotactic MRI was then obtained for the treatment planning. Pre- and postcontrast thin-slice (1-mm) volume acquisition axial and coronal MRI sequences were obtained. When an MRI was not able to be obtained due to medical contraindications (e.g., a cardiac pacemaker), a thin-slice stereotactic CT scan was obtained with and without contrast administration. Radiosurgical dose plans were formulated under the direction of a neurosurgeon in conjunction with a medical physicist and radiation oncologist. In general, the maximum dose to the optic apparatus was kept to 8 Gy or less. The Leksell Gamma Unit Model U was used until July 2001 when the C model (Elekta Instruments, Inc.) was instituted. The Gamma Knife Perfexion unit has been used to treat patients since 2007. The Kula software was used for dose planning from 1990 to June 1994, and then subsequently Elekta’s Gamma Plan software was used. All patients in this series were treated with single-session radiosurgery. Initial tumor volume was assessed by contouring and then using Gamma Plan software. Gamma Knife Radiosurgery Parameters The mean dose to the tumor margin was 13.5 ± 3.5 Gy (range 4.8–30 Gy) with a mean maximal dose of 31 ± 5 Gy (range 23–40 Gy). Only 1 patient received a marginal dose of less than 9 Gy, and the 5th to 95th percentiles for marginal dose ranged from 9 to 20 Gy. The majority of tumors were treated with multiple isocenters (mean 13 ± 10 isocenters, range 1–43 isocenters) to the mean prescription isodose line of 42% ± 9% (range 30%–55%). Radiosurgery parameters are detailed in Table 2.
Radiosurgery of large skull base meningiomas
TABLE 1. Baseline characteristics of 75 patients with large skull base meningiomas treated with GKRS Characteristic
Value (%)
Females Age (yrs) Median ± SD Range Previous resection Previous radiotherapy Previous GKRS Initial presentation* Headache Subjective dizziness Seizure CN III/IV/VI V VII VIII IX/X XI XII Hypopituitarism Body weakness Dysgraphia Cognition Location Parasellar Petroclival Clival Tentorial CPA Petrous Sphenoidal Jugular foramen Tumor volume (cm3) Mean ± SD Range Tumor diameter (cm) Mean maximum Range Edema on pre-GKRS imaging
51 (68) 55 ± 13 19–85 45 (60) 4 (5) 3 (4) 18 (28) 14 (22) 3 (5) 29 (45) 33 (52) 9 (14) 11 (17) 1 (2) 0 (0) 1 (2) 2 (3) 4 (6) 2 (3) 1 (2) 42 (56) 9 (12) 6 (8) 3 (4) 10 (13) 2 (3) 2 (3) 1 (1) 14.1 ± 6.7 8.1–54.8 3.8 ± 0.8 2.5–6 4 (5)
* Clinical presentation data available for 64 patients.
Clinical and Radiological Follow-Up Patients were routinely followed clinically and radiologically every 6 months for the first year, annually until 5 years after radiosurgery, and then every 2 years thereafter. At each follow-up visit, a neurological examination was performed to evaluate for new neurological deficits, and neuroimaging studies were reviewed to assess tumor response by both a neurosurgeon and neuroradiologist. All
TABLE 2. GKRS treatment parameters Characteristic
Mean ± SD (range)
Margin dose (Gy) Maximum dose (Gy) Isodose line (%) No. of isocenters
13.5 ± 3.5 (4.8–30) 31 ± 5 (23–40) 42 ± 9 (30–55) 13 ± 10 (1–43)
follow-up evaluation was performed at the University of Virginia, unless the patient was unable to travel to our institution. In such cases, the referring physician performed the follow-up and provided documentation of the patient’s neurological status, as well as follow-up imaging. All neuroimaging studies were reviewed by a neurosurgeon and a neuroradiologist at the University of Virginia. Lesions were categorized into the following locations defined by the presumed origin of maximal volume: sellar or parasellar, CPA, petrous, clival, petroclival, jugular foramen, and sphenoid.2,3,5,50,54,55 Petroclival meningiomas were defined as tumors whose maximal volume was centered over the region between the petrous apex and the upper two-thirds of the clivus.3,50 All parasellar lesions were in close proximity to the sella with cavernous sinus invasion.54 Imaging outcome was determined by the last available radiological examination by a neurosurgeon and a neuroradiologist.47 A decrease or increase in tumor size was defined as a 15% or greater change in tumor volume as compared with the volume at the time of GKRS.47 Tumors with less than a 15% change from their initial volume were considered radiologically stable. Any patient with tumor progression of greater than 15% was considered a treatment failure even if this stabilized with further GKRS or surgery.47 To make the determination of tumor volume, the tumor was outlined on radiological images and serial volumetric calculations were performed using Image J in all patient imaging studies (NIH).47 The presence of perilesional edema was defined as FLAIR or T2-weighted changes around the tumor. These changes were evaluated by a neuroradiologist (M.S.) who assessed the radiosurgical planning MR images and the follow-up MR images for the presence of baseline edema and an exacerbation of the same after radiosurgery. Statistical Analysis Data are presented as median or mean and range for continuous variables, and as frequency and percentage for categorical variables. Calculations of normality were performed by ladder of powers and assessed graphically. Statistical analyses of categorical variables were conducted using chi-square, Fisher’s exact, and Mantel-Haenszel tests for linear association as appropriate. Statistics of means were performed using an unpaired Student t-test, both with and without equal variance (Levene’s test) as necessary and Wilcoxon rank-sum tests when variables were not normally distributed. The following dependent variables were assessed in univariate and multivariate analysis: tumor-free progression, worsening or new deJ Neurosurg Volume 122 • February 2015
365
R. M. Starke et al.
cline in neurological function, and favorable outcome (no tumor progression or worsening or new decline in neurological function). Kaplan-Meier risk of tumor progression was calculated. Factors predictive of tumor progression (p < 0.20)4 were entered into multivariate Cox regression analysis to assess hazard ratios (HRs). Clinical covariates predicting new or worsening decline in neurological function with a univariate p value < 0.20 were included in multivariable logistic regression analysis. Additionally, clinical covariates predicting unfavorable outcome with a univariate p value < 0.20 were included in multivariable logistic regression analysis. Clinically significant variables and interaction expansion covariates were further assessed in both Cox and logistic multivariable analysis as deemed relevant. Youden indices were calculated to determine cutoffs for the dichotomized continuous variable tumor volume that yielded the optimal discrimination of tumor progression and overall outcome. Those p values ≤ 0.05 were considered statistically significant.
Results
Clinical Outcomes Detailed clinical follow-up data were available for 64 patients. Neurological function was unchanged in 37 patients (58%), improved in 16 patients (25%), and declined in 11 patients (17%). Specific alterations in neurological function are displayed in Table 3. The most common CNs
to have new or worsening function following GKRS were II, III, and V, which each occurred in 4 patients (6%). Additionally, the most likely CNs to demonstrate improvement following GKRS were III, V, and VII occurring in 3 (5%), 4 (6%), and 3 (5%) of patients, respectively. In the current series, there was no evidence of vascular injury or brainstem ischemia as a result of SRS. Predictors of new or worsening decline in neurological function in univariate analysis are demonstrated in Table 4. In multivariate analysis, the factors predictive of new or worsening decline in neurological function were history of surgery (OR = 3.00, 95% CI 1.13–7.95; p = 0.027), presentation with any CN deficit from III to VI (OR = 3.94, 95% CI 1.49–10.24; p = 0.007), and decreasing maximal dose (OR = 0.76, 95% CI 0.63–0.93; p = 0.007). Tumor progression was present in 7 (64%) of 11 patients with new or worsening decline in neurological function. Radiological Outcome The mean follow-up duration was 6.5 years (range 0.5– 21 years). During this time, 37 patients (49%) displayed no change in tumor volume, 26 (35%) had decreased volume, and 12 (16%) displayed increased volume. Kaplan-Meier analysis demonstrated radiological progression-free survival at 3, 5, and 10 years to be 90.3%, 88.6%, and 77.2%, respectively (Fig. 1). Four patients had a history of edema prior to SRS (5.3%). Following radiosurgery, 4 patients had new or
TABLE 3. Neurological outcomes for 64 patients with large skull base meningiomas treated with GKRS Variable
Deficit Before GKRS
Total patients (%) Total no. of deficits per patient (%) Dizziness Body weakness Seizure Hypopituitarism Dysgraphia Cognition CN deficit* CN I CN II CN III CN IV CN V CN VI CN VII CN VIII CN IX CN X CN XI CN XII
59 (92)* 14 (22) 4 (6) 3 (5) 2 (3) 2 (3) 1 (2) 52 (81) 2 (3) 19 (30) 24 (38) 3 (5) 33 (52) 16 (25) 9 (14) 11 (17) 0 1 (2) 0 1 (2)
* Some patients had multiple deficits † Includes patients who did not present with a deficit. 366
J Neurosurg Volume 122 • February 2015
Stable Deficit
Improved Deficit
Worsened or New Deficit
37 (58)†
16 (25)
11 (17)
12 (19) 3 (5) 3 (5) 2 (3) 2 (3) 0 33 (52) 2 (3) 15 (23) 18 (28) 3 (5) 26 (41) 16 (25) 6 (9) 10 (16) 0 0 0 1 (2)
2 (3) 1 (2) 0 0 0 1 (2) 12 (19) 0 1 (2) 3 (5) 0 4 (6) 0 3 (5) 1 (2) 0 1 (2) 0 0
0 0 1 (2) 0 0 1 (2) 9 (14) 0 4 (6) 4 (6) 1 (2) 4 (6) 1 (2) 0 0 0 0 0 0
Radiosurgery of large skull base meningiomas
TABLE 4. Predictors of new or worsening neurological deficits Predictive Factor Univariate analysis* Location All other locations Parasellar/petroclival History of radiotherapy History of surgery Any CN deficit from III to VI Seizure Increasing volume Decreasing peripheral dose Decreasing maximal dose Multivariate analysis History of surgery Any CN deficit from III to VI Decreasing maximal dose
OR
95% CI
p Value
1.00 7.10 6.25 2.40 2.98 11.56 1.06 0.82 0.83
0.85–59.54 1.07–36.54 1.08–5.34 1.43–6.23 0.95–141.13 0.98–1.15 0.64–1.05 0.71–0.96
0.071 0.042 0.032 0.004 0.055 0.171 0.118 0.012
3.00 3.94 0.76
1.13–7.95 1.49–10.24 0.63–0.93
0.027 0.007 0.007
* Factors predictive of new/worsening symptoms (p < 0.20).
worsening edema within 18 months of treatment: 1 patient had progression of preexisting edema and 3 patients developed edema following SRS. The edema improved in 1 patient following a short course of steroids but persisted in the other 3 patients. Three patients with preexisting edema experienced improvement in the edema following radiosurgery. There were no cases of malignant transformation of an existing meningioma or radiation-induced secondary tumor development. Factors predictive of tumor progression in univariate analysis are displayed in Table 5. Prior resection was not predictive of tumor control (Fig. 2). Regarding tumor control, Youden indices demonstrated that the optimal breakpoint for tumor volume occurred at 14 cm3 (Fig. 3). In the Cox multivariable analysis, the covariates associated with tumor progression included presentation with any CN deficit from III to VI (HR = 3.78, 95% CI 1.91–7.45; p < 0.001), history of prior radiation therapy (HR = 12.06, 95% CI 2.04–71.27; p = 0.006), and tumor volume greater than 14 cm3 (HR = 6.86, 95% CI 0.88–53.36; p = 0.066; Table 5; Figs. 4 and 5). Further Treatments Following SRS During follow-up, 2 patients underwent fractionated radiation therapy for out-of-field tumor progression (3%), and 5 patients underwent resection for tumor progression (7%). All tumors resected after radiosurgery were confirmed to be WHO Grade I meningiomas. Three patients had evidence of radiological progression of ventriculomegaly, but none of these patients required a CSF diversion procedure. Overall Outcome After SRS Unfavorable outcome, defined as tumor progression and/or new or worsening decline in neurological function, was observed in 12 patients (18.8%), compared with 52 patients (81.2%) who had a favorable outcome. Of those
Fig. 1. Graph of tumor-free progression after GKRS.
with an unfavorable outcome, 7 patients (10.9%) exhibited both tumor progression and neurological decline. Univariate predictors of unfavorable overall outcome (i.e., tumor growth and/or neurological decline) are detailed in Table 6. Statistically significant factors in univariate analysis for an unfavorable outcome included petroclival/parasellar location, prior radiotherapy, prior surgery, any CN deficit from III to VI, diplopia on presentation, preexisting seizures, pre-GKRS edema, and decreasing maximal and peripheral dose (Table 6). Patients with a history of resection before radiosurgery were significantly more likely to have an unfavorable outcome in univariate analysis (OR = 2.37, 95% CI 1.08–5.21; p = 0.032), but prior resection was not an independent predictor of outcome. Youden indices did not provide a significant cutoff for the dichotomized continuous variable “tumor volume” that yielded the optimal discrimination of overall outcome. In the multivariate analysis, the factors predictive of unfavorable outcome included history of prior radiation TABLE 5. Predictors of tumor progression Predictive Factor Univariate analysis* Location All other locations Parasellar/petroclival History of radiotherapy History of hydrocephalus Any CN deficit from III to VI Diplopia on presentation Seizure Tumor volume >14 cm3 Decreasing maximal dose Multivariate analysis History of radiotherapy Any CN deficit from III to VI Tumor volume >14 cm3
HR
95% CI
p Value
1.00 5.11 3.73 1.19 3.29 2.47 7.64 6.86 0.93
0.65–39.91 1.00–13.94 1.00–1.41 1.79–6.05 0.79–7.78 1.53–38.04 0.88–53.36 0.85–1.02
0.120 0.050 0.056
